Search Constraints

Start Over
Filtering by: Creator Lim, Ho-Jin Remove constraint Creator: Lim, Ho-Jin Advisor Thomas, David R. Remove constraint Advisor: Thomas, David R.

Search Results

  • Performance of some tests for bivariate independence, interchangeability, and equality of distributions of ratios

    Creator:
    Ferguson, Nancy Lo
    Abstract:
    The purpose of this thesis is to derive and compare asymptotic performances of certain tests of independence, interchangeability and equality of the distributions of_ratios Y/X for positive variates and Y with joint c.d.f. H(x,y) of particular forms. Three bivariate families, Morgenstern, Plackett, and Moran, are considered which permit marginal c.d.f.'s...
    Resource Type:
    Dissertation
    Full Text:
    ----> co under the following conditions: (1) xlimco xiF(xly) - F(x)] = 0 , (2) lim y [H-FG] = 0

  • Non parametric c-sample tests for arbitrarily censored data

    Creator:
    Hirunraks, Anek
    Abstract:
    Consider the c- sample testing problem with null hypothesis H[subscript]o : F₁(x) = ... = F[subscript]c(x), where F[subscript]j(x), j = 1 ..., c, represent absolutely continuous distribution functions for failure distributions. Furthermore, consider arbitrarily (right) censored samples, which may arise from putting the (censored) experimental units on test at different...
    Resource Type:
    Dissertation
    Full Text:
    problem with null hypothesis Ho : F 1(x) = = F (x), where F.(x), j = 1, , c, represent absolutely

  • Goodness of fit tests under random censorship

    Creator:
    Chen, Jiin-chiang
    Abstract:
    The classical chi-square goodness of fit test and the Neyman smooth tests are generalized for arbitrarily right-censored samples. The test statistics are shown to have limiting chi-square distributions under completely specified (simple) null hypotheses and random censorship. Limiting distributions of the generalized chi-square goodness of fit tests are also established...
    Resource Type:
    Dissertation
    Full Text:
    ) E Rk. Suppose E(e ) tlYn exists for all n in he neighborhood of t = 0, lim E(e n'00 exists, ane Rn

  • Inference procedures for pairs of distributions with proportional failure rate functions

    Creator:
    McRae, Kenneth Bruce
    Abstract:
    Some nonparametric maximum likelihood estimation procedures are developed for the class of pairs of distributions which have proportional failure rate functions. Special consideration is given to the case in which the shape of the failure rate functions are assumed to be either increasing or decreasing. Estimators of the proportionality constant,...
    Resource Type:
    Dissertation
    Full Text:
    4)(0), we consider the limiting values of the component functions; and nlimg(0) = lim E5=0m. 1 [1

  • Robust estimation for the mean of skewed distributions

    Creator:
    Hussein, Osama Abdelaziz
    Abstract:
    Common estimators of the mean g(θ) = Jx dFθ(x) in skewed distribution models may be sensitive to contamination by a few large observations. It is then desirable to consider robust estimators g(θ) .The approach of Hampel (1968), who defines an estimator θ for the parameter vector θ to be optimal...
    Resource Type:
    Dissertation
    Full Text:
    F is defined as IF(x;T,F) := Lim ((T[(1-0F+e-45 ] - T(F))/el e-60 x provided that the limit exists

  • Efficiency comparisons of goodness-of-fit tests for Weibull and gamma distributions with singly censored data

    Creator:
    Bargal, Adel Ibrahim
    Abstract:
    Generalizations of the smooth goodness-of-fit test statistics, the components of Cramer-von Mises type statistics and the Pearson chi-square statistics are considered for censored samples. Asymptotic efficiency comparisons are made between these general goodness-of-fit tests for Weibull and gamma distributions for certain parametric alternatives in the case of singly censored samples....
    Resource Type:
    Dissertation
    Full Text:
    ), Ho Weibull (2) vs HA :Weibull (3). 6 Efficiencies of the smooth test ip (X0) relative to the

  • Confidence bands for percentiles in the linear regression model

    Creator:
    Thomas, Dana Lester
    Resource Type:
    Dissertation
    Full Text:
    implies (ww) lim (z + Q.) 0, so (Z,W') (1 + :4)1/2 (ii) If (Z,W') e J and Z > 0 then by the Cauchy